Abstract
Abstract
Objectives
To develop and internally validate leakage-robust landmark models to estimate 12-month fracture-related infection (FRI) risk after open tibial fracture at admission and postoperative day 3 (POD3).
Methods
In this single-center retrospective cohort, adults with open tibial fractures treated with limb salvage between January 2023 and June 2024 were included. Definite 12-month FRI was adjudicated using the international FRI consensus definition. Two landmark models were prespecified: admission and postoperative day 3 (POD3). Predictors included injury severity, contamination, host and physiologic factors, antibiotic timing, and, at POD3, early clinical course, routine laboratory results, debridement timing, and soft-tissue coverage status. Elastic net–penalized logistic regression with restricted cubic splines was used. Missing data were addressed with multiple imputation nested within 1,000-bootstrap internal validation. Performance was assessed by AUC, Brier score, calibration, and decision curve analysis.
Results
Among 913 patients, 125 (13.7 %) developed FRI. The admission model achieved an AUC of 0.77 and Brier score of 0.097. The POD3 model improved performance (AUC 0.82; Brier 0.089) and showed slightly greater net benefit across 5–25 % thresholds.
Conclusions
Landmark models provided internally validated 12-month FRI risk estimates at admission and POD3, with modest improvement after POD3 updating. External validation and recalibration are required to evaluate its potential clinical deployment.